One cannot probably fully understand how the anti-Majority Text position has come to dominate Textual Criticism, without tracing a little bit of its history.
Although cautions against simply 'counting MSS' began to be vocalized as early as Wetstein (1700s), truly articulate arguments probably begin with Tregelles, Tischendorf and Alford. These men first advanced the argument in favor of the oldest manuscripts. Reasoning against the Majority Text was merely a corollary, but a necessary step to clear a path for consideration of the oldest texts. So early arguments against the the Majority Text were crude and usually minimal in content.
One of the better popular arguments was offered by T.S. Green, in his A Course of Developed Criticism, (London, 1856) Intro. Here for the first time we get a window into the thinking behind the rather rash and easy rejection of the traditional (majority) text:
"The work to which the critic of the NT is called, must consist to a considerable extent in disentangling the text from intrusive and usurping matter, having its origin in the margin ; in detaching accretions, and replacing whatever may have been dislodged by a spurious rival : and with this view one leading principle must be especially noticed. (1)
Corruption of this particular kind must be the work of time, because the growth of such matter itself would be gradual, (2) and its sliding into the text by the agency of reckless, ill-taught, and foolish hands, and through the general propensity of copyists for amplification, (3) would be likewise gradual : the evil, too, unchecked in its earlier stages by due watchfulness or control, would go on spreading with the advance of time. (4)
It follows of necessity from this, that the more ancient documents will in general exhibit a greater approach to purity in this particular respect than those of later date, and, as a practical consequence, that the adverse testimony of but a few witnesses of high antiquity, in the case of matter of questioned genuineness, must receive the first and foremost regard, even though it were certain that their text was unsound in certain other respects, as, for instance, in the touches of critical hands. (5)
Fewness must not discourage a reliance on their testimony, because, if an intrusion took place at a particular point at a remote date and there is sufficient proof that such mischief was very early at work such a numerical disparity is precisely the state of things to be encountered in the body of surviving documents, where the really ancient must, from the very nature of things, form but a small minority, and even of these all cannot be expected to have escaped intrusive influence. (6)
This canon, as it may be called, does not rest on an unreasoning prepossession in favour of antiquity, but is a logical consequence from unquestionable premises.
Since in citing the MSS which exhibit a certain reading, a great preponderance of mere numbers is imposing in appearance, and may seem to be a circumstance that cannot lightly be set aside or countervailed by other considerations, it will be well to state fairly and precisely how much may be concluded from the circumstance:
Out of the entire body of existing copies, as has already been remarked, those of high antiquity form a very small portion ; and, accordingly, any great majority of the whole must be almost entirely composed of those of later date. Whenever, therefore, a particular reading is supported by a greatly preponderating part of the mass in contrast with a group of distinctively ancient copies, all that can be at once concluded from this bare fact is, that the reading in question had a settled currency in later times. This narrow conclusion is all that in such a case can be taken into account from MSS. alone in a discussion of the claims of a reading; without any prejudice, however, to arguments for antiquity and genuineness which may be derivable from other quarters notwithstanding. (7)
In one particular way mere numbers would be important evidence of genuineness, namely, in case there were something in the character of the reading itself adverse to its acceptance in the presence of rivals, and, therefore, to that currency which those numbers indicate. (8)
Mere numerical considerations do not therefore possess that prime importance which they might at first sight seem to claim, and which they have too frequently been allowed to exercise.
Unfortunately, this basic argument, as reasonable as it sounds at a glance, has several fatal flaws, which we will analyze shortly.In a review of authorities special regard will reasonably be paid to antiquity: but this must not be overstrained into a summary neglect of more recent witnesses, as necessarily offering nothing worthy of notice.
The critic should not suffer himself to be encumbered by prepossessions or assumptions, nor bind himself to the routine of a mechanical method of procedure. If he allows himself to be thus warped and trammelled, instead of ever maintaining the free employment of a watchful, calm, and unfettered mind, he abandons his duty and mars his work. "
(1) It is remarkable that Green's sketch poses that the main problems 'to a considerable extent' consist of insertions, substitutions, and absorbed marginal glosses. He skips over completely the other glaringly common error, accidental omissions by homoeoteleuton (h.t.). Its as if these errors, though they comprise a large chunk of significant variants in both individual manuscripts and text-types, were an insignificant aspect of a critic's work. That Green is fully aware of the existance of h.t. errors, is clear from other statements.
Hort likewise gave lip-service to h.t. errors, while failing to discuss the possibility of same among the hundreds of probable omissions he adopted. The unspoken assumption is that the antiquity of a reading somehow makes it exempt from being an h.t. error, even though h.t. errors are rampant in even the most ancient of manuscripts. That assumption is blatantly false however, and this casts all such omissions in the critical text under grave suspicion.
Error Rate vs. Error Accumulation Rate:
(2) Green says the accumulation of errors would be gradual. This much is true, in that errors will occur at different places in the text, and in all ages. Thus errors will be spread through the text and across time fairly randomly and evenly.
But Green and all early critics fail to distinguish between the raw error-rate, and the rate of accumulation of errors. Because there is no awareness of the difference between these two rates, there is no awareness of the independence of these two rates.
For example, say the error-rate is 1 error per verse in each copy. Without error-checking, a 3rd generation copy would have about 3 errors per verse. We could estimate the error accumulation rate at about 1 error per verse per generation.
But now suppose each copy is proof-read, and about half the errors are caught. The raw error-rate is the same, but the error accumulation rate is now only 1 error per verse for every two generations, or half that of the raw error rate.
A simple checking procedure has drastically altered the accumulation rate, without any change in scribal skill or raw error rate.
Suppose now that about every 10 generations, a copy is checked against a much older copy (10 generations earlier). Again half the errors are caught. But this time, ALL the errors accumulated over 10 generations will be open to checking. The new copy, which would have had 10 errors per verse, now has about 5 per verse (given the same quality of proof-reading as before). The 10th generation corrected copy is the equivalent of a 5th generation copy. Again the error accumulation rate has been halved, with a simple, occasional procedure.
But more importantly, the same reduction in errors was accomplished this time by the same procedure performed only 1/5 as often! Checking a much older copy is potentially an order of magnitude more effective than using the same technique on every copy in every generation!
While the Tortoise can certainly achieve the same purity with thorough hard work, the Hare can effectively skip doing any error-checking at all for many generations, and yet achieve the very same purity, with only the same mediocre skill and effort case by case as the tortoise.
This shows us a few revealing things:
(a) The raw error-rate is effectively independent of the error accumulation rate.Simple proof-reading methods effectively telescope generational copying effects, reducing real generations to a fraction of their potential handicap for practical purposes. These methods can even be inconsistent and sporadically applied, with the very same results in quality control.
(b) The error accumulation rate can be drastically attenuated by simple techniques, without needing to increase scribal skills.
(c) Most importantly, longer-term procedures can effectively bypass accumulated errors and virtually 'reset' the error count to near-zero repeatedly, causing the error accumulation rate to slow to a virtual crawl.
(3) Green mentions "the general propensity of copyists for amplification". That is, he assumes copyists were more likely to add than to omit material, emphasizing Griesbach's canon, 'Prefer the Shorter Reading'. This was not however a general tendency, and the rule could only be applied in a restricted set of situations with caution. For a modern critique of this idea, see Royse.
Such opinions in Green's time were based on impressions and guesstimates; but modern studies have made clear that scribes were far more often prone to omissions than additions. This has major consequences for the credibility of 19th century critical reconstructions of the text.
(4) Green's claim about the continual spread (accumulation) of errors is also based on a failure to grasp the dramatic impact of error-correction using earlier copies. We have explored above what even occasional use of early manuscripts can do. But mixture effectively demolishes skeptical arguments regarding generational corruption. Significantly, the majority of 'mixture' (cross-correction of a MS with another text) will occur using older manuscripts as a guide. This has a devastating effect on the accumulation of errors, effectively wiping out whole clusters of error and continually purging the text.
(5) Green's argument here is that his previous observations inevitably lead to the idea that a few older manuscripts will outweigh a large number of later manuscripts. But this is a serious non-sequitor. His model assumes there is no significant error-checking, proof-reading, and so the age of a manuscript also indicates its genealogical relationship in regard to an accumulating error-pool.
But this is an absurdity, given that the majority of copies cannot be placed in a sequential copying line. Rather, they are individual copies of independent lines of transmission reaching back to early times. The diversity of both texts, geographical origins, and community heritage of these texts indicates they are not direct descendants of the earliest surviving manuscripts, or even close relatives in a common copying stream.
Furthermore, some copies clearly 'jump the que', being direct copies of much earlier texts which are independent of surviving older copies.
Since there is no way to intelligently group late manuscripts all together in contrast to early ones, we must deal with each manuscript or small family independently. They cannot be crudely separated from early copies, placed together and de-valued en masse.
(6) Now Green imagines that an 'early interpolation' enhances his argument that a few early MSS would have the true reading against a mass of late ones. For if the interpolation were early enough to be the landslide majority reading in the great remainder of later MSS, this would automatically place the alteration earlier in time than the relatively late 4th century date of the extant Uncials. But if so, then the question of whether it is an omission in the Uncials or an addition in the later MSS must be settled on other grounds. The 4th century Uncials are just as secondary to the early corruption of the text as other lines of transmission.
It would be more convincing if the omission were found in a significant number of independent later MSS as well as a few early Uncials. Then the addition would clearly be 'late' relative to the Uncials. Such nonsense has the appearance of reasonableness but is really just another case of naive, simplistic, oversimplification so rampant in the 19th century.
(7) Green's argument here runs essentially thus: (a) By the nature of the case, there will always be a large number of later MSS, and few early ones, therefore (b) a majority reading by itself only means it was settled in later times.
The argument here is again a non-sequitur and false. When the mass of later MSS are closely examined, it becomes clear that they represent such a wide diversity of transmission-lines, and variety of ages of for their texts, that any overwhelming agreement must drive the reading back to the 4th century and earlier, contemporary with the surviving most ancient MSS.
The false assumption of relative lateness of text for an overwhelming majority reading found in the mass of later MSS must be dismissed. But we were never in the predicament that this would be the only available evidence in any case. Most majority readings can also be found to be supported by early versions, and most importantly early quotations from the large numbers of early Christian writers (ECWs). The claim that 'later MSS always outnumber early ones, therefore their testimony is worthless' is a ridiculous argument.
The correct conclusion is, the mass of later evidence must be carefully considered alongside earlier evidence, especially since early evidence is fragmentary and unreliable.
(8) Now Green suggests that (internal) evidence of the difficulty of, or hostility to a reading comes into play to give added weight to a majority reading. Thus for Green and friends, a majority reading can be valued if its somehow difficult, adverse, or offensive.
We need hardly comment on the subjectivity of this 'exception' that Green proposes to reinstate a majority reading to respectability. In comparison to textual evidence, all such personal conjecture and opinion must be secondary, and considered on its own merits, but hardly at all, next to overwhelming textual evidence in the form of landslide majority support.